U-2 spy plane caused widespread shutdown of U.S. flights | Hacker News Reader

U-2 spy plane caused widespread shutdown of U.S. flights

reuters.com

93 pointshappyscrappy48 comments

Kind of a misleading headline.

U-2s have been flying over Southern California at over 60k feet for 50 years (that is, after all, where they're based), and there are other more modern planes that also routinely fly at that altitude.

If we're pointing fingers, the new ERAM system that was just installed (in December, I believe) in the LA ATC is a more likely culprit, although even that isn't likely the whole story; this can't be the first high flying plane in the last 6 months. So, more likely an edge case that took 6 months to crop up, or maybe a new bug caused by a recent patch. But as the article notes, the FAA isn't talking, so we still don't know.

"U-2 spy plane caused widespread shutdown of U.S. flights" is a cool headline, but an accurate one would be "We still aren't sure why new computer system caused widespread shutdown of U.S. flights".

ERAM is a quite troubled project. Just read the Inspector General's report. Sadly, many decades after the Mythical Man Month and much learned about software project management, we continue to see such efforts flounder.

http://www.oig.dot.gov/sites/dot/files/ERAM%20Final%20Report...

The ERAM report is sad. I guess if Uncle Sam pays when you oveerrun your budget, it is a bonus.

If a U2 can set its own altitude on the transponder the other possibility here is that the flight operator set it to an incorrect altitude that appeared it was flying in the same airspace as commercial planes but not within standard corridors.

This would generate significant number of collision warnings, given how busy LA regional airspace is, and that in turn might have demonstrated a limit to the number of simultaneous collion events this new ATX system can handle.

That sounds plausible

I wonder if it's because can fly higher than 65,535ft.

"the maximum altitude transmitted by a U-2 transponder is FL600: even if the aircraft is flying well above it, the Dragon Lady’s mode C will show no higher than 60,000 feet." Which might be the issue if there was an off by one error in the code.

I want to know why a spy plane has a transponder. Can it be turned off when flying over Cuba?

A spy plane has a transponder so you don't shoot a friendly/neutral aircraft down (or waste time classifying it as friendly/neutral.)

The codes it squawks can be modified in various ways but there will be scenarios when you want your transponder, such as flying in an area where you have air superiority.

I think the idea of the U-2, which was developed before stealth was really a thing, is that although you can see it, it flies so high it is out of range of your interceptors.

After Gary Powers, they probably double-check that the countries they fly it over don't have modern AA capabilities.

Even commercial airliners' transponders can be trivially switched off in the cockpit.

Because it's a legal requirement to fly in a lot of airspace, to ensure that you show up on radar, and it's far easier to install one than to wrangle an exception. Of course it can be turned off, just like any other piece of equipment.

> far easier to install one than to wrangle an exception

You are talking about an agency that builds and operates planes that nobody else knows about. They can have any "exception" they want by just shrugging and saying the aircraft in question does not even exist.

All military aircraft have transponders because the air bases have ATC as well and (at least in the US) are near commercial airports they have to work with to avoid collisions during takeoff and landing.

Flying a military airplane through civilian airspace without participating in civilian ATC sounds like an excellent way to end one's career in the military.

Sure, they can just say the aircraft does not exist, if they stick to their own space. A U-2 is not going to do that.

Even the B-2 bomber had a retractable radar reflector to defeat it's stealth. When flying in civilian airspace the reflector would extend so ATC could track the plane. Standard stuff to prevent collisions.

Why would they have a retractable reflector rather than simply using a standard radar transponder? Sounds like BS.

To start with, a reflector is passive, so it can't fail like a transponder. Failing is a problem when flying an intentionally very hard to see/detect airframe in dense civilian airways. Transponders only provide a secondary radar hit, while many (older) systems are looking for a primary hit to locate aircraft, which a reflector would do.

Before calling BS you could just Google it too. ;) For example, here is a nice explanation of the F117's reflector. http://en.wikipedia.org/wiki/Lockheed_F-117_Nighthawk

Any system that relies on a primary hit is going to have trouble. Something as simple as a composite glider is not going to show up well[1]. I'm not aware of any ATC system that relies on primary returns.

As for Googling it, that's an interesting picture. It's not retractable, but I suppose the point is made. Perhaps the B-2's is retractable. I did Google it this time and found nothing but some random airliners.net forum posts, not exactly reliable.

[1] http://www.dg-flugzeugbau.de/radar-reflector-e.html

Nice. So the system depends entirely on transponder?

What will happen when one plane is at the right altitude but report one higher by human or machine error on a day where everyone is flying blind? ...not that we have any blind day in LA but still...

It is standard procedure for ATC to request altitude (or, more commonly, to get unsolicited altitude reports) on initial contact, and to not trust the Mode C data until it can be cross-checked for exactly this reason.

That crossed my mind; I really hope this isn't the case, because that would mean a lot of people who should be expected to know these sorts of things with such familiarity that it's like their lives depended on it, actually don't.

Now thanks to 2048, if it did turn out to be 16-bit overflow, maybe quite a lot more of the general public would understand this too...

Seems unlikely; altitudes are normally managed in multiples of 100, e.g. FL350 = 35,000 ft. But of course it depends on the system's internal data structures.

Whoosh!!!

More details, and more expert analysis, available at the Aviationist[1].

[1] http://theaviationist.com/2014/05/03/u-2-freezes-lax-radar/

Also see previous discussion: https://news.ycombinator.com/item?id=7691583

Given the number of new/identified aircraft spotted in the last few months there's a strong possibility this was simply not a U-2.

http://theaviationist.com/2014/04/23/two-different-black-pro...

While that is a possibility, there's nothing to suggest it is stronger than the reported id of the plane, so I'll stick with U-2.

I just watched the U-2 documentary on that site, and I love that among all the super advanced sensors, they are still using a yaw string! :-)

Sometimes the simplest solution is still the best :)

Quick sea story. To avoid interfering with launch and recovery operations, there is a 3-mile radius cylinder (the "stack") around U.S. aircraft carriers that military aircraft have to get permission from the tower to enter. One time, while operating off the coast of Pakistan, a U-2 called us up and asked permission to enter the stack, "at Angels Sixty."

Or in other words, 10 miles directly above all the other aircraft. They got permission.

Nice. That story reminded me of the good ol' SR-71 speed check one [1]. I'm sure I'm not alone in this among my fellow HN 'Aviation Week' enthusiasts:

[1] https://groups.google.com/forum/m/#!topic/rec.aviation.stori...

Note that the failure was in a new-ish, recently-deployed air traffic control system named ERAM (En Route Automation Modernization), and its backup system:

https://en.wikipedia.org/wiki/ERAM

Do they fuzz systems like this, or do they just ship it as soon as the test suite passes?

It seems like there's a whole world of hurt waiting to arrive if people can start to inject data into these systems.

So, what variable max values were not accounted for and tested in the software?

Could it be that the U2 has sending an ADS-B signal with incorrect altitude to hide its actual location?

I would think that would be a feature of such a plane and it may have been activated by mistake.

U2s aren't particularly stealthy and if you want to hide the location you just turn the transponder off.

I captured ADS-B packets from a U-2 in the LAX area (using an RTL-SDR dongle) on 2013-06-18. If you want to see them for yourself, the decoded packets (hex) are at https://dl.dropboxusercontent.com/u/7443278/Mode%20S%20Captu... (along with other aircraft).

The U-2 I picked up is ICAO ae094b, registration 68-10336[1]. You can decode the packet information using dump1090[2]:

  ./dump1090 --net-only > out.txt

Then in another terminal:

  cat mode-s-capture-2013-06-18.raw | nc localhost 30001

Then in out.text, look for "ae094b". You'll see packets like

  *0206042b1365a5;
  CRC: ae094b (ok)
  DF 0: Short Air-Air Surveillance.
    VS             : Airborne
    CC             : 1
    SL             : 0
    Altitude       : 60000 feet
    ICAO Address   : ae094b

[1] http://www.airliners.net/search/photo.search?regsearch=68-10... [2] https://github.com/MalcolmRobb/dump1090

Altitude (appears) out of bounds, therefore we assume it is at an unknown altitude, so could be at any altitude, therefore any plane on a collision bearing regardless of altitude must be diverted.

Wonder why nobody posted this on Sunday Bloody Sunday. Still, today is a Beautiful Day for a drone story.

Protip: don't fly above 65535 ft. Or book a flight with ' ; -- drop tables: air

She moves in mysterious ways.

I sincerely doubt the accuracy of this article. The U-2 has been around for 50 years and only now has an ATC system been "overwhelmed" by it?

Is the system at LAX new?

Yes, that's what it says in the article.

The idea that a flight routing system wouldn't take into consideration the altitude of the planes involved seems a bit like a movie rather then reality.

Now if they said the problem was the U-2 was moving so quickly that the calculations of all the planes in the air ground the systems to a halt maybe that would make sense.

Something smells very Hollywood here with this story.

U-2s don't fly very fast (~500mph). For comparison, a 747 cruises at about 570mph.

Others have made the point that perhaps it exceeded the system's MAX_ALTITUDE

I'm no longer surprised by errors in software, even critical applications like air traffic control and monitoring. The contracts go to lowest bidder, not the people that actually have the capability of engineering the system properly.

Reply on news.ycombinator.com